Differential contractility regulates cancer stem cell migration.

Cancer stem cells (CSCs) are known to have a high capacity for tumor initiation and the formation of metastases. We have previously shown that in collagen constructs mimetic of aligned extracellular matrix architectures observed in carcinomas, breast CSCs demonstrate enhanced directional and total motility compared with more differentiated carcinoma populations. Here, we show that CSCs maintain increased motility in diverse environments including on 2D elastic polyacrylamide gels of various stiffness, 3D randomly oriented collagen matrices, and ectopic cerebral slices representative of a common metastatic site. A consistent twofold increase of CSC motility across platforms suggests a general shift in cell migration mechanics between well-differentiated carcinoma cells and their stem-like counterparts. To further elucidate the source of differences in migration, we demonstrate that CSCs are less contractile than the whole population (WP) and develop fewer and smaller focal adhesions and show that enhanced CSC migration can be tuned via contractile forces. The WP can be shifted to a CSC-like migratory phenotype using partial myosin II inhibition. Inversely, CSCs can be shifted to a less migratory WP-like phenotype using microtubule-destabilizing drugs that increase contractility or by directly enhancing contractile forces. This work begins to reveal the mechanistic differences driving CSC migration and raises important implications regarding the potentially disparate effects of microtubule-targeting agents on the motility of different cell populations.

Senescence Defines a Distinct Subset of Myofibroblasts That Orchestrates Immunosuppression in Pancreatic Cancer.

Pancreatic ductal adenocarcinoma (PDAC) therapeutic resistance is largely attributed to a unique tumor microenvironment embedded with an abundance of cancer-associated fibroblasts (CAF). Distinct CAF populations were recently identified, but the phenotypic drivers and specific impact of CAF heterogeneity remain unclear. In this study, we identify a subpopulation of senescent myofibroblastic CAFs (SenCAF) in mouse and human PDAC. These SenCAFs are a phenotypically distinct subset of myofibroblastic CAFs that localize near tumor ducts and accumulate with PDAC progression. To assess the impact of endogenous SenCAFs in PDAC, we used an LSL-KRASG12D;p53flox;p48-CRE;INK-ATTAC (KPPC-IA) mouse model of spontaneous PDAC with inducible senescent cell depletion. Depletion of senescent stromal cells in genetic and pharmacologic PDAC models relieved immune suppression by macrophages, delayed tumor progression, and increased responsiveness to chemotherapy. Collectively, our findings demonstrate that SenCAFs promote PDAC progression and immune cell dysfunction. Significance: CAF heterogeneity in PDAC remains poorly understood. In this study, we identify a novel subpopulation of senescent CAFs that promotes PDAC progression and immunosuppression. Targeting CAF senescence in combination therapies could increase tumor vulnerability to chemo or immunotherapy. See related article by Ye et al., p. 1302.

Ultra-low frequency EPR using longitudinal detection and fictitious-field modulation.

When viewed in a rotating frame of reference, a transverse-plane radiofrequency (RF) field manifests as a longitudinal field component called the fictitious field. By modulating the RF field and thus the fictitious field, detectable longitudinal magnetization patterns have previously been shown to be measurable. By combining fictitious-field modulation and longitudinal detection, here we demonstrate EPR spectroscopy and one-dimensional imaging in a custom-built longitudinal detection system operating at an ultra-low frequency (24 MHz) for detecting electron spins with short (~nanoseconds) relaxation times. Simultaneous transmit and receive with low transmitter leakage level (~80 dB isolation) is also demonstrated.

Quadratic Autoencoder (Q-AE) for Low-Dose CT Denoising.

Inspired by complexity and diversity of biological neurons, our group proposed quadratic neurons by replacing the inner product in current artificial neurons with a quadratic operation on input data, thereby enhancing the capability of an individual neuron. Along this direction, we are motivated to evaluate the power of quadratic neurons in popular network architectures, simulating human-like learning in the form of "quadratic-neuron-based deep learning". Our prior theoretical studies have shown important merits of quadratic neurons and networks in representation, efficiency, and interpretability. In this paper, we use quadratic neurons to construct an encoder-decoder structure, referred as the quadratic autoencoder, and apply it to low-dose CT denoising. The experimental results on the Mayo low-dose CT dataset demonstrate the utility and robustness of quadratic autoencoder in terms of image denoising and model efficiency. To our best knowledge, this is the first time that the deep learning approach is implemented with a new type of neurons and demonstrates a significant potential in the medical imaging field.

A two-dimensional feasibility study of deep learning-based feature detection and characterization directly from CT sinograms.

Machine Learning, especially deep learning, has been used in typical x-ray computed tomography (CT) applications, including image reconstruction, image enhancement, image domain feature detection and image domain feature characterization. To our knowledge, this is the first study on machine learning for feature detection and analysis directly based on CT projection data. Specifically, we present neural network methods for blood vessel detection and characterization in the sinogram domain avoiding any partial volume, beam hardening, or motion artifacts introduced during reconstruction. First, we estimate sinogram domain vessel maps using a residual encoder-decoder convolutional neural network (REDCNN). Next, we estimate the vessel centerline and we extract the vessel-only sinogram from the original sinogram, eliminating any background information. Finally, we use a fully connected neural network to estimate the vessel lumen cross-sectional area from the vessel-only sinogram. We trained and tested the proposed methods using CatSim simulations, real CT measurements of vessel phantoms, and clinical data from the NIH CT image database. We achieved encouraging initial results showing the feasibility of CT analysis in the sinogram domain. In principle, sinogram domain analysis should be possible for many other and more complicated clinical CT analysis tasks. Further studies are needed for this sinogram domain analysis approach to become practical for clinical applications.